#nvidiagpus

[govindhtech]

Amazon Braket Notebooks support CUDA-Q NVIDIA for HQC

New Amazon Braket Notebook Instances Natively Support NVIDIA CUDA-Q, Revolutionising Quantum Development on AWS

An important quantum computing platform upgrade from AWS is that Amazon Braket Notebook settings now natively support CUDA-Q, NVIDIA's open-source hybrid quantum-classical computing architecture. Increasing NVIDIA's accessibility streamlines hybrid quantum-classical system developers' and scholars' workflows.

The connection lets customers run CUDA-Q programs in Amazon Braket Jupyter notebook instances without configuration. Native support is enabled by upgrading laptop instances to Amazon Linux 2023, which improves security, performance, and compatibility for quantum development workflows.

Smooth Hybrid Workflow Development

With this intrinsic feature, quantum researchers and developers can simply create and test hybrid quantum-classical algorithms. The platform simplifies hybrid algorithm creation, modelling, and execution in Braket-managed notebooks.

Native CUDA-Q support provides several key features:

GPU-accelerated calculations and quantum simulations can be combined in hybrid workflows. GPU acceleration: Amazon Braket makes NVIDIA GPUs available to developers for faster algorithm creation and simulation.

Access to Quantum Hardware: The platform lets users simply switch from simulation to Braket's quantum hardware. IonQ, Rigetti, and IQM offer single-managed Quantum Processing Units (QPUs). Amazon This release strengthens Braket's position as a comprehensive quantum computing research and development platform. It simplifies CUDA-Q use in managed notebooks by eliminating the need for developers to control local deployment or route execution via Hybrid Jobs.

New Environment Access and Pre-Installed Packages When customers create an Amazon Braket notebook instance, they get the latest kernel. New conda_braket tile may appear when you open the notebook. By selecting this tile under “Notebook” or “Console,” the new conda_braket kernel launches a Jupyter notebook or Python console session.

“CUDA-Q and Braket” tile opens example notebooks using the conda_braket kernel as the default kernel.

Braket notebooks come with the latest compatible packages for the four leading quantum development frameworks Braket, CUDA-Q, PennyLane, and Qiskit. Developers can verify installed packages with a command.

Several critical package versions are confirmed in the output, including:

Amazon Basket Algorithm Library 1.6.2

amazon-braket-default-simulator=1.31.4

Amazon-braket-pennylane-plugin=1.33.5

Amazon-braket-schemas=1.26.1

Amazon-braket-sdk=1.102.6

cudaq=0.12.0.post1

cudaq-qec=0.4.0.post1

cudaq-solvers=0.4.0

PennyLane=0.42.3

PennyLane_Lightning=0.42.0

1.4.4 qiskit

0.17.2 qiskit-aer

Qiskit-algorithms=0.4.0

qiskit-ionq=0.6.1

0.6.0 qiskit_braket_provider

Amazon Braket notebook instances come with quantum applications using Braket, CUDA-Q, PennyLane, and Qiskit. Users can view CUDA-Q examples on the Launcher page by clicking “CUDA-Q and Braket”. This opens the notebook 0_hello_cudaq_jobs.ipynb and shows more examples under nvidia_cuda_q/.

How to Implement and Examples

Small, CPU-based instance types like ml.t3.medium are recommended for laptop instances by AWS. Amazon Braket Hybrid Jobs can perform GPU-intensive CUDA-Q apps. GPU-powered instances like ml.p3.8xlarge are recommended for these intensive tasks. GPU-accelerated instances include:

ml.p3.2xlarge (1 NVIDIA V100 GPU).

ml.p3.8xlarge (4 V100 GPUs).

ml.p3.16xlarge (8 V100 GPUs).

When running GPU-powered instances concurrently through Hybrid Jobs, a customer's Amazon Elastic Compute Cloud (EC2) service quota should exceed the number of instances they want.

The Previous Docker-Based Setup

Developers needed a Jupyter kernel running a CUDA-Q Docker container to execute CUDA-Q apps interactively in Braket notebooks before this native integration. This enables customers to run CUDA-Q simulators on Amazon Braket-supported quantum hardware backends and powerful NVIDIA GPUs using the open-source NVIDIA CUDA-Q platform and Amazon Braket Hybrid Jobs.

Braket Notebook instances come with Docker, allowing controlled development using NVIDIA NGC Container Registry images.

The previous approach required several steps to configure a custom kernel:

Dockerfile creation: Stable Dockerfiles require CUDA-Q images like nvcr.io/nvidia/quantum/cuda-quantum:cu12-0.9.1. This file explained installing ipython, ipykernel, and amazon-braket-sdk.

Next, the image was built in three to five minutes.

For the new kernel (such as docker_cudaq), a kernel.json file was created containing command-line options for starting the kernel with the newly created Docker image.

GPU Access: Add “–gpus=all” to the kernel to access GPU instances like ml.p3.2xlarge, ml.p3.8xlarge, and ml.p3.16xlarge in CUDA-Q program.configuration in JSON.

This configuration mounts /home/ec2-user/amazon-braket-examples/examples inside the CUDA-Q container. If they needed more Python modules or updated CUDA-Q versions, they merely needed to tweak the Dockerfile and rebuild the Docker image.

Standard development in a controlled environment with native support now avoids this complicated setup.

[bestwebhosting-6]

Why Live Cam Changes Everything For AI Companions

Late one night I wasn't really in the mood to type. I just wanted company. That's when SweetDream's live cam feature stopped being a novelty for me and started feeling like a small gift. Available with select characters on sweetdream.ai, a live cam session turns your AI companion from a chat window into something closer to a face you're happy to see.

Here's why it works so well: SweetDream nails the foundation. The conversations are remarkably natural and emotionally tuned in, remembering context across days so nothing feels reset. The AI-generated photos and videos are stunning, the voice calls genuinely sound human, and video calls are there too. Live cam just extends that warmth into a live, shared moment.

And because privacy is treated as non-negotiable, you can relax into it. Everything you share and everything you create stays discreet and yours. That peace of mind is part of why SweetDream feels like the best AI girlfriend platform out there — it lets the connection be the only thing you're thinking about.

[bestwebhosting-6]

GPU Architecture

NVIDIA’s GPU architecture combines multi-core processors, advanced memory, and specialized cores like Tensor and RT, delivering high performance for AI, gaming, and computing. Its scalability and efficiency make it a key driver of innovation in modern data centers, scientific research, and next-gen applications.

[govindhtech]

Master Robotics Development with NVIDIA Isaac Sim

NVIDIA Isaac Sim

NVIDIA and university experts are working together to prepare robots for surgery.

Researchers from the University of Toronto, UC Berkeley, ETH Zurich, Georgia Tech, and NVIDIA created ORBIT-Surgical, a simulation framework for training robots that could improve surgical teams’ abilities while lightening the cognitive load on surgeons.

It supports over a dozen manoeuvres, including grabbing small objects like needles, transporting them from one arm to another, and precisely putting them, that are modelled after the training curriculum for laparoscopic procedures, also known as minimally invasive surgery.

NVIDIA Isaac Sim, a robotics simulation platform for creating, honing, and testing AI-based robots, was used to build the physics-based framework. In order to allow photorealistic rendering, the researchers used NVIDIA Omniverse, a platform for creating and implementing sophisticated 3D applications and pipelines based on Universal Scene Description (OpenUSD), totrain imitation and reinforcement learning algorithms on NVIDIA GPUs.

At this week’s IEEE International Conference on Robotics and Automation (ICRA) in Yokohama, Japan, Today, ORBIT-Surgical will be presented. You may now get the open-source code package on GitHub.

AI’s Stitch Saves Nine

The foundation of ORBIT-Surgical is Isaac Orbit, a modular robot learning framework based on Isaac Sim. Several libraries for imitation learning and reinforcement learning which educate AI agents to imitate real-world expert examples are supported by Orbit.

Da Vinci Research Kit

With the surgical framework, developers may use reinforcement learning and imitation learning frameworks running on NVIDIA RTX GPUs to train robots, such as the da Vinci Research Kit robot, or dVRK, to manipulate both hard and soft things.

More than a dozen benchmark activities are introduced by ORBIT-Surgical for surgical training, including one-handed tasks like raising a suture needle to a precise location, placing a shunt into a blood vessel, and picking up a piece of gauze. It also includes two-handed activities such as putting a threaded needle through a ring pole, passing a needle between two arms, and stretching two arms to designated locations while avoiding obstructions.

In comparison to current surgical frameworks, the team is able to increase robot learning time by an order of magnitude by creating a surgical simulator that leverages GPU acceleration and parallelization. On a single NVIDIA RTX GPU, they discovered that the robot digital twin could be trained to perform activities like lifting a suture needle and inserting a shunt in less than two hours.

Because rendering in Omniverse provides visual realism, ORBIT-Surgical enables researchers to produce high-fidelity synthetic data that may be used to train AI models for perceptual tasks like identifying surgical instruments in real-world footage taken in operating rooms.

The team’s proof of concept demonstrated how integrating simulation and actual data enhanced an AI model’s ability to distinguish surgical needles from photos, hence lowering the requirement for big, costly real-world datasets for model training.

ICRA

Papers at ICRA authored by NVIDIA

Many will be talking about geometric fabrics at the IEEE International Conference on Robotics and Automation (ICRA), which is being held in Yokohama, Japan, from May 13–17. One of the seven publications that members of the NVIDIA Robotics Research Lab and their collaborators have submitted addresses that topic.

Geometric fabrics: what are they?

Trained policies in robotics are inherently approximative. While they generally act appropriately, on occasion they may move the robot too quickly, bump into objects, or jerk it around. One may never be certain of what might happen.

Therefore, when trained policies, particularly those trained through reinforcement learning, are applied to a physical robot, a layer of low-level controllers is used to intercept the policy’s commands. After that, they convert the orders to meet the hardware’s constraints.

Run those controllers with the policy while you’re training RL policies. The researchers discovered that vectorizing those controllers to make them available for use during training and deployment was a special feature that could be provided with their GPU-accelerated RL training tools.This study aims for that.

Humanoid robot businesses, for instance, might showcase demonstrations of their devices using low-level controllers that not only maintain the robot’s balance but also prevent it from running its arms into itself.

The controllers that were vectorized by the researchers come from their previous work on geometric fabrics. At the previous year’s ICRA, the paper Geometric Fabrics: Generalising Classical Mechanics to Capture the Physics of Behaviour, took home the best paper prize.

Surgical Orbit

NVIDIA Isaac Sim powers photorealistic rendering in ORBIT-Surgical, a physics-based surgical robot simulation framework running on the NVIDIA Omniverse platform.

In order to train imitation learning and reinforcement learning algorithms that support the research of robot learning to supplement human surgical skills, GPU parallelization is utilised. It also makes it possible to generate synthetic data for active perception tasks that is realistic. The researchers show off the sim-to-real transfer of learnt policies onto a real dVRK robot utilising ORBIT-Surgical.

Upon publication, the underlying ORBIT-Surgical robotics simulation application will be made available as a free, open-source package.

Visit ICRA to meet NVIDIA Robotics partners

At ICRA, NVIDIA robotics partners are showcasing their most recent innovations.

Based in Zürich Introducing ANYmal Research, a whole software suite from ANYbotics that gives customers access to low-level controls all the way down to the ROS system. Hundreds of academics from leading robotics research institutes, including as the University of Oxford, ETH Zürich, and the AI Institute, are part of the community known as ANYmal Research. (Handle IC010)

Munich-based Franka Robotics showcases their work using the Franka toolkit for Matlab and NVIDIA Isaac Manipulator, an AI companion powered by NVIDIA Jetson that powers robot control. (Handle IC050)

Read more on govindhtech.com

[govindhtech]

Innovative AI-Ready Precision Workstations with NVIDIA GPUs

AI-Ready Precision Workstations Performance

A combination of AI-ready Precision workstations and RTX-accelerated AI-development tools is a great method for software developers to get a head start on the production and use of artificial intelligence applications. This combination is both rapid and straightforward. In light of the recent announcements made by NVIDIA, which aim to facilitate the adoption of large language models (LLMs) by offering enhanced performance and efficient processing of AI-ready Precision workstations, they are overjoyed to make these announcements. By using generative artificial intelligence, these improvements make it possible for software developers to construct unique apps and services.

Earlier this week, NVIDIA made the announcement that enhancements for the Gemma family of models are now available across all NVIDIA AI platforms. Among the most recent additions to Google’s open model portfolio, these models are state-of-the-art lightweight open language models with two billion and seven billion parameters. They are capable of running on a wide variety of platforms, ranging from Dell Precision AI-ready Precision workstations to Dell scalable AI server infrastructure.

By highlighting the strategic use of TensorRT runtime to boost the performance of various latent learning models (LLMs), such as Stable Diffusion XL (SDXL) Turbo and latent consistency models, which are two widely preferred approaches for speeding Stable Diffusion, they are contributing to the forefront of innovation. Furthermore, TensorRT-LLM has been used for the purpose of RTX-acceleration of text-based models, this includes Llama 2, Mistral, and Phi-2.

Bringing these models to you is the entryway to the possibilities that are provided by the entire AI portfolio that Dell provides. Developers are provided with a wide variety of tools to make their artificial intelligence projects more successful by Dell. Whether it’s taking use of cutting-edge AI development tools, choosing powerful workstations, or venturing into the realm of LLMs, Dell’s dedication to innovation ensures that developers have the resources they need to confidently traverse their path through the world of artificial intelligence.

With the TensorRT engines operating on the most intelligent, secure, and manageable commercial PCs in the world, Dell with NVIDIA provides a solid foundation for accelerated prototyping and exploration in the rapidly expanding world of artificial intelligence. This foundation enables one to easily access moderately complex development pipelines, which are typically difficult to enable from scratch.

A retrieval-augmented-generation (RAG) tool known as Chat with RTX and the NVIDIA AI Enterprise software that is presently in shipping are two examples of the plethora of tools that are available to developers via NVIDIA’s extensive ecosystem. With the future AI Workbench, developers will be able to quickly construct, collaborate on, and repeat generative artificial intelligence and data science projects. This is something that they can look forward to very much.

A few clicks are all that is required for developers to scale up their work from a local workstation or RTX PC to the cloud or a data center. The beta version of AI Workbench is not yet available. The NeMo framework, NVIDIA RAPIDS, TensorRT, and TensorRT-LLM are all integrated with GPU-accelerated development software via these tools. This allows you to fine-tune these LLMs and deploy them for your particular use case.

This program is a terrific example of what is possible when artificial intelligence is applied to your data while being protected by your firewall. It is a quick way to achieve higher levels of productivity, business insight, and efficiency. To be successful in the rapidly developing field of artificial intelligence, it is essential to have access to the appropriate tools. The combination of speed, dependability, and scalability that Dell AI-ready Precision workstations for artificial intelligence development provide is unrivaled.

These workstations are accelerated with NVIDIA RTX GPUs. By delivering end-to-end AI solutions and services that are tailored to meet clients wherever they are in their AI journey, Dell Technologies provides the world’s biggest portfolio of artificial intelligence (AI) solutions, ranging from desktop to data center to cloud.

FAQS:

AI-ready Precision workstations are what exactly?

In addition to having strong CPUs, enough memory, and NVIDIA RTX GPUs, these high-performance workstations are built expressly for AI applications that are particularly demanding.

The use of AI-ready Precision workstations that are equipped with NVIDIA GPUs brings around what advantages?

Performance that is noticeably faster: RTX graphics processing units (GPUs) speed up artificial intelligence operations like as training, inference, and data analysis, which results in both greater productivity and speedier results. Increased dependability and stability: Precision workstations are designed for professional usage, which guarantees a smooth operation and little downtime during tasks that are of crucial importance. Capacity to accommodate future requirements: These workstations are capable of being customized with a wide range of components in order to accommodate the increasing processing requirements.

For the creation of artificial intelligence, which software packages are provided or suggested on these workstations?

The NeMo framework, NVIDIA RAPIDSTM, TensorRT, and TensorRT-LLM are some of the NVIDIA RTX-accelerated artificial intelligence development tools that Dell makes available to its customers. These technologies assist in the creation and deployment of AI applications in an effective manner.

Read more on Govindhtech.com